Semester

Spring

Date of Graduation

2022

Document Type

Dissertation

Degree Type

PhD

College

Statler College of Engineering and Mineral Resources

Department

Civil and Environmental Engineering

Committee Chair

Omar I. Abdul-Aziz

Committee Co-Chair

Radhey. S. Sharma

Committee Member

Radhey. S. Sharma

Committee Member

Seung Ho Hong

Committee Member

Debangsu Bhattacharyya

Committee Member

Yasin Yilmaz

Abstract

Dissolved oxygen (DO) is a general indicator of stream water quality and ecosystem health. However, the concentration of in-stream DO is controlled by various climatic, land use/cover, hydrologic, biochemical, and ecological drivers. Observational data for stream DO are often unavailable at the desired temporal and spatial scales. Accurate and reliable prediction of stream DO concentration based on a small set of environmental drivers is important to guide the water managers and policymakers to achieve and maintain healthy streams. This dissertation focused on identifying the dominant controls of stream DO and understanding the mechanisms and environmental regimes by employing machine learning, orthogonal data analytics, and similitude (parametric reduction) analyses with dimensionless numbers. Parsimonious models were then developed to robustly predict DO across inland and coastal streams of the contiguous USA. This research utilized available data of monthly to quarterly sampling frequencies during 1998-2015 for 86 inland and 117 coastal streams, representing gradients in climate, hydrology, biogeochemistry, and land use/cover in USA. In general, water temperature (Tw), pH, and total phosphorous (TP) represented the dominant climatic and biogeochemical controls of DO in both inland and coastal streams. Tw had a strong and dictating control on stream DO, representing mainly the dissolution process of oxygen in water and partly the catalytic effect of temperature on stream metabolism. pH showed moderate positive linkage with stream DO, representing the concurrent outcomes of in-stream metabolic respiration. DO was also notably controlled by TP, which drives stream metabolism as the limiting nutrient with typically abundant total nitrogen (TN). Emergent power-law scaling models were developed to robustly predict DO (mg/l) based on only Tw (Kelvin) and pH, with the exponents of ~ 15/2 and ½ (respectively) across the inland and coastal streams of USA (Nash-Sutcliffe Efficiency, NSE = 0.73-0.83). Furthermore, two sets of mechanistically meaningful dimensionless numbers were obtained by involving stream DO and the environmental variables. The dimensionless DO/TP number demonstrated a robust and predictive scaling law with the Redfield ratio TN/TP and hydro-bioclimatic number (Pa/(TP.Tw.cp)) with the exponents of ¹/5 and ⅔, respectively across the various streams (NSE = 0.80-0.81); where cp is the specific heat of water and Pa is the atmospheric pressure. The DO saturation number (DO/DOsat) scaled with the stream metabolic number (Pa3.W/(Tw3.cp3.TP.H.S.D)) with a robust power law exponent of ⅓ (NSE = 0.79-0.80); where S is the stream salinity, H is the hydrogen ion concentrations, W is the stream width, and D is the stream depth. The research findings provided new insights, fundamental knowledge, and robust predictive tools to achieve and maintain good water quality and ecosystem health in streams across the contiguous U.S. and beyond.

Embargo Reason

Publication Pending

Available for download on Thursday, May 01, 2025

Share

COinS